Amazon's Alexa endorsed Kamala Harris while refusing to say a single positive word about Donald Trump.
That was 2024 – and Big Tech called it an "error."
Now Google's AI just did something that makes Alexa look like a typo.
Google Gemini Political Bias Exposed: Republicans Only, Zero Democrats
Wynton Hall – Breitbart's social media director and author of the new book Code Red: The Left, the Right, China, and the Race to Control AI – ran a simple test through Google Gemini Pro's deep research function: which U.S. senators have made statements violating its hate speech policies?
The answer came back with Republicans only.
Zero Democrats.
Not one.
Sen. Marsha Blackburn of Tennessee made the list for characterizing transgender identity as a harmful cultural "influence" and using the word "woke." Sen. Tom Cotton of Arkansas got flagged for cosponsoring legislation to exclude transgender students from sports.
Gemini found nothing objectionable from a single Democratic senator.
Not Rep. Dan Goldman, who warned on national television that Trump needed to be "eliminated." Not Texas Democratic House candidate Rep. Jolanda Jones, who made a throat-slashing gesture on CNN while explaining how to deal with political opponents.
Fox News Digital reviewed Hall's screen recording of the prompt and findings. Google did not respond to requests for comment.
This is what AI censorship looks like before the platform admits it – a machine quietly deciding whose speech is dangerous, teaching hundreds of millions of users that one party's mainstream positions are hateful and the other party's calls for violence are just politics.
Silicon Valley Bias: Why 85% of Big Tech Donations Go to Democrats
This didn't happen by accident.
Hall's research shows 85% of political donations from employees at Apple, Meta, Amazon, and Google go to Democrats.
That's not a coincidence. That's a company culture – and that culture gets baked directly into the AI those employees build and fine-tune. PayPal co-founder Peter Thiel said it plainly: "Silicon Valley is a one-party state."
Nobody writes a memo saying flag Republicans and protect Democrats. They don't have to. When an entire workforce shares the same worldview, the AI reflects it – the same way a child learns by absorbing what the adults around him treat as normal.
JD Vance saw this coming. "If you go back to 2020, 2021, Google search was so biased in the left-wing direction that I think it actually changed America's political system," he told Fox News. "I don't want that to happen with artificial intelligence."
It already has.
How AI Censorship Starts: Big Tech Trains the Machine on Liberal Media
Google's AI wasn't trained on random internet data.
It was trained heavily on The New York Times, The Atlantic, Reuters – outlets that treat conservative positions on gender and the border as fringe, dangerous, or both. Conservative publications were largely excluded from those training sets.
So the machine absorbs the mainstream media's assumptions, packages them as neutral fact, and delivers answers to users who have no reason to question an encyclopedia.
Think about your neighbor – the one who doesn't follow politics closely, who just wants a straight answer. She asks Google Gemini which senators are making dangerous statements. She gets a Republican-only list, presented with the authority of a research tool, with no indication it reflects the politics of the engineers who built it.
She doesn't scroll past it like a biased headline. She believes it. Then she tells her book club.
Hall calls it "Big Tech's consolidating control on steroids." He's right. Shadow banning was bad. This is shadow banning with a research degree and a neutral voice.
How to Fight Back Against Big Tech AI Censorship
Hall's solution in Code Red is direct: demand full transparency in AI training data and cut every taxpayer-funded contract with vendors whose systems show political bias.
Sen. Lindsey Graham showed in 2024 that pressure works – he put Amazon "on notice" over the Alexa incident and the company scrambled to fix it within days.
Google just told every American which senators its machine considers dangerous. Neighbors, grandchildren, voters who trust it like a textbook – they're being taught the same answer every time they ask a question.
"Whoever wins the AI fairness battle," Hall writes, "will shape the minds and political attitudes of future generations."
Google is already playing. The question is whether conservatives show up.
Sources:
- Andrew Murray, "Google Gemini declares only GOP senators violate hate speech policy, zero Democrats, author claims," Fox News, March 16, 2026.
- Wynton Hall, Code Red: The Left, the Right, China, and the Race to Control AI, HarperCollins, March 17, 2026.
- Danielle Wallace, "Amazon Alexa gives starkly different answers when asked why to vote for Trump versus Kamala Harris," Fox Business, September 3, 2024.
- Julia Johnson, "Lindsey Graham puts Amazon 'on notice' over Alexa's potential election interference," Fox News, September 4, 2024.
- Lucas Nolan, "New York Post Publishes Stunning Excerpt of 'CODE RED,'" Breitbart, March 15, 2026.
